import matplotlib.pyplot as plt
import pandas as pd
import numpy as np
from bokeh.plotting import figure, show
from bokeh.models import ColumnDataSource, HoverTool,FactorRange
from bokeh.io import output_notebook
import seaborn as sns
from bokeh.transform import dodge
from bokeh.palettes import Category20

from sklearn.model_selection import train_test_split,cross_val_score, KFold
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import roc_curve, auc,roc_auc_score
from sklearn.preprocessing import LabelEncoder

from bokeh.resources  import settings

settings.resources = 'inline'

# data loading

data = pd.read_csv("HR_comma_sep.csv")

data.head()

data.isnull().sum()

satisfaction_level       0
last_evaluation          0
number_project           0
average_montly_hours     0
time_spend_company       0
Work_accident            0
left                     0
promotion_last_5years    0
sales                    0
salary                   0
dtype: int64

# Renaming the 'sales' column to 'department'
data = data.rename(columns={"sales": "department"})
data.columns

Index(['satisfaction_level', 'last_evaluation', 'number_project',
       'average_montly_hours', 'time_spend_company', 'Work_accident', 'left',
       'promotion_last_5years', 'department', 'salary'],
      dtype='object')

data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 14999 entries, 0 to 14998
Data columns (total 10 columns):
 #   Column                 Non-Null Count  Dtype  
---  ------                 --------------  -----  
 0   satisfaction_level     14999 non-null  float64
 1   last_evaluation        14999 non-null  float64
 2   number_project         14999 non-null  int64  
 3   average_montly_hours   14999 non-null  int64  
 4   time_spend_company     14999 non-null  int64  
 5   Work_accident          14999 non-null  int64  
 6   left                   14999 non-null  int64  
 7   promotion_last_5years  14999 non-null  int64  
 8   department             14999 non-null  object 
 9   salary                 14999 non-null  object 
dtypes: float64(2), int64(6), object(2)
memory usage: 1.1+ MB

data.describe(include=['O'])

# Convert 'salary' and 'department' to categorical data type
data['salary'] = data['salary'].astype('category')
data['department'] = data['department'].astype('category')

salary_dict = dict(enumerate(data['salary'].cat.categories))
department_dict = dict(enumerate(data['department'].cat.categories))

# Create mapping dictionaries
salary_mapping = dict(enumerate(data['salary'].cat.categories))
salary_mapping_reverse = {v: k for k, v in salary_mapping.items()}

department_mapping = dict(enumerate(data['department'].cat.categories))
department_mapping_reverse = {v: k for k, v in department_mapping.items()}

# Create new columns for numerical values
data['salary_num'] = data['salary'].cat.codes
data['department_num'] = data['department'].cat.codes

data.head()

# Exclude categorical columns for mean calculations
columns_to_exclude = ['department', 'salary']
numeric_df = data.drop(columns=columns_to_exclude)

# Group by the 'left' column
left_summary = numeric_df.groupby(by=['left'])
numeric_df.mean()

satisfaction_level         0.612834
last_evaluation            0.716102
number_project             3.803054
average_montly_hours     201.050337
time_spend_company         3.498233
Work_accident              0.144610
left                       0.238083
promotion_last_5years      0.021268
salary_num                 1.347290
department_num             5.870525
dtype: float64

numeric_df.describe()

# Count the number of employees who stayed and left
left_count = numeric_df['left'].value_counts().reset_index(name="count")

# Plotting a pie chart to show the turnover ratio
labels = ['Stayed', 'Left']
sizes = left_count['count']
colors = ['#1f77b4', '#ff7f0e']  # Blue and orange colors
explode = (0, 0.1)  # Explode the 'Left' slice

plt.figure(figsize=(6, 6))
plt.pie(sizes, labels=labels, colors=colors, explode=explode,
        autopct='%1.1f%%', shadow=True, startangle=140)
plt.title("Employee Turnover Ratio")
plt.axis('equal')  # Equal aspect ratio ensures that pie is drawn as a circle.
plt.show()

# Calculate average satisfaction level based on years spent at the company
time_mean_satisfaction = numeric_df.groupby('time_spend_company')['satisfaction_level'].mean().reset_index(name="average_satisfaction")

# Plotting a bar chart to show satisfaction vs. years spent at the company
plt.figure(figsize=(10, 6))
plt.bar(time_mean_satisfaction['time_spend_company'], time_mean_satisfaction['average_satisfaction'], color='#1f77b4')
plt.title("Employee Satisfaction vs. Years Spent at Company")
plt.xlabel("Years at Company")
plt.ylabel("Average Satisfaction")
plt.grid(True)
plt.show()

# Convert the crosstab to a format suitable for Bokeh
depart_left_table = pd.crosstab(index=numeric_df['department_num'], columns=numeric_df['left'])
depart_left_table.reset_index(inplace=True)

# Map numerical department values to categorical names
departments = [department_dict[dept] for dept in depart_left_table['department_num']]
stayed = depart_left_table[0].tolist()
left = depart_left_table[1].tolist()

data_left = {'departments': departments,
        'Stayed': stayed,
        'Left': left}

source = ColumnDataSource(data=data_left)

# Create the figure
p = figure(y_range=FactorRange(*departments), height=400, width=800,
           title="Employee Turnover by Department", toolbar_location=None, tools="")

# Add bars
p.hbar(y=dodge('departments', -0.15, range=p.y_range), right='Stayed', height=0.3, source=source,
       color=Category20[20][0], legend_label="Stayed")
p.hbar(y=dodge('departments', 0.15, range=p.y_range), right='Left', height=0.3, source=source,
       color=Category20[20][1], legend_label="Left")

# Customize the plot
p.y_range.range_padding = 0.1
p.ygrid.grid_line_color = None
p.xaxis.axis_label = "Count"
p.yaxis.axis_label = "Department"
p.legend.orientation = "horizontal"
p.legend.location = "top_center"

# Output to notebook
output_notebook()
show(p)

# Crosstab to compare salary distribution by department
depart_salary_table = pd.crosstab(index=data['department'], columns=data['salary'])

# Ensure all salary levels are present for each department
depart_salary_table = depart_salary_table.reindex(columns=salary_dict.values(), fill_value=0)

# Convert the Crosstab data to Bokeh-compatible format
departments = depart_salary_table.index.tolist()
low_salary = depart_salary_table[salary_dict[0]].tolist()
medium_salary = depart_salary_table[salary_dict[1]].tolist()
high_salary = depart_salary_table[salary_dict[2]].tolist()

source = ColumnDataSource(data={
    'departments': departments,
    'Low Salary': low_salary,
    'Medium Salary': medium_salary,
    'High Salary': high_salary
})

# Create the figure
p = figure(x_range=departments, height=400, width=800,
           title="Employee Salary Levels by Department", toolbar_location=None, tools="")

# Add vertical bars with hover tooltips
p.vbar_stack(stackers=['Low Salary', 'Medium Salary', 'High Salary'], x='departments', width=0.9, color=['#1f77b4', '#ff7f0e', '#2ca02c'],
             source=source, legend_label=['Low Salary', 'Medium Salary', 'High Salary'])

# Hover tooltip configuration
hover = HoverTool()
hover.tooltips = [
    ('Department', '@departments'),
    ('Low Salary', '@{Low Salary}'),
    ('Medium Salary', '@{Medium Salary}'),
    ('High Salary', '@{High Salary}')
]
p.add_tools(hover)

# Customize the plot
p.y_range.start = 0
p.xgrid.grid_line_color = None
p.axis.minor_tick_line_color = None
p.outline_line_color = None
p.legend.location = "top_right"
p.legend.orientation = "vertical"

# Show plot
output_notebook()
show(p)

# Crosstab to compare turnover by salary level
salary_left_table = pd.crosstab(index=data['salary'], columns=data['left'])

# Plotting bar chart to show turnover rate by salary level
fig, ax = plt.subplots(figsize=(6, 5))
bars = salary_left_table.plot(kind='bar', stacked=False, ax=ax, color=['#1f77b4', '#ff7f0e'])

# Setting legend labels
ax.legend(['Stay', 'Left'])
ax.set_title("Impact of Employee Salary on Turnover")
ax.set_xlabel("Salary")
ax.set_ylabel("")
plt.xticks(rotation=0)

# Adding labels on top of each bar
for bar in bars.patches:
    # Get the height of the bar
    height = bar.get_height()
    # Add text annotation above each bar with its value
    ax.annotate(f'{height}', xy=(bar.get_x() + bar.get_width() / 2, height),
                xytext=(0, 3),  # 3 points vertical offset
                textcoords="offset points",
                ha='center', va='bottom')

plt.show()

# Crosstab to compare turnover by promotion status
promotion_left_table = pd.crosstab(index=data['promotion_last_5years'], columns=data['left'])
promotion_dict = {0: "No Promotion", 1: "Promotion"}

# Plotting bar chart to show turnover rate by promotion status
fig, ax = plt.subplots(figsize=(6, 6))
promotion_left_table.plot(kind='bar', stacked=False, ax=ax, color=['#1f77b4', '#ff7f0e'])
ax.legend(['Stay', 'Left'])
ax.set_title("Impact of Employee Promotions in Last 5 Years on Turnover")
ax.set_xlabel("Promotion Status")
ax.set_ylabel("Count")
ax.set_xticklabels([promotion_dict[val] for val in promotion_left_table.index], rotation=0)
plt.show()

evaluation_left_table = pd.crosstab(index=data['last_evaluation'], columns=data['left'])

# Plotting bar chart to show turnover rate by performance evaluation
fig, ax = plt.subplots(figsize=(10, 5))
evaluation_left_table.plot(kind='bar', stacked=False, ax=ax, color=['#1f77b4', '#ff7f0e'])
ax.legend(['Stay', 'Left'])
ax.set_title("Impact of Employee Performance Evaluation on Turnover")
ax.set_xlabel("Performance Evaluation")
ax.set_ylabel("Count")
plt.xticks(rotation=90)
plt.show()

corr = numeric_df.corr()
mask = np.zeros_like(corr)
mask[np.tril_indices_from(mask)]=True

with sns.axes_style("white"):
    sns.set(rc={'figure.figsize':(7,7)})
    ax = sns.heatmap(corr,
                xticklabels=True, yticklabels=True,
                cmap='RdBu',
                mask=mask,
                fmt='.3f',
                annot=True,
                linewidths=.5,
                vmax=.4,
                square = True
                # center = 0
                )

plt.title("Correlation")
label_x = ax.get_xticklabels()
plt.setp(label_x,rotation=45, horizontalalignment='right')
plt.show()

# Selecting relevant features and target variable
features = ['satisfaction_level', 'last_evaluation', 'number_project',
            'average_montly_hours', 'time_spend_company', 'Work_accident',
            'promotion_last_5years']
target = 'left'

X = data[features]
y = data[target]

# Splitting the data into training and testing sets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42)

# Creating the RandomForestClassifier
rdf = RandomForestClassifier(n_estimators=100, random_state=42)
cv=KFold(n_splits=10, shuffle = True, random_state=90)
rdf_score=cross_val_score(rdf,X_train,y_train,cv=cv).mean()
# Training the model
rdf.fit(X_train, y_train)

# Predicting probabilities on the test set
y_pred_prob = rdf.predict_proba(X_test)[:, 1]
rdf_y_pred = rdf.predict(X_test)
# Computing ROC curve and AUC
fpr, tpr, thresholds = roc_curve(y_test, y_pred_prob)
roc_auc = auc(fpr, tpr)
rdf_auc = roc_auc_score(y_test,y_pred_prob)

# Plotting ROC curve
plt.figure(figsize=(6, 6))
plt.plot(fpr, tpr, color='blue', lw=2, label='ROC curve (AUC = %0.2f)' % roc_auc)
plt.plot([0, 1], [0, 1], color='gray', linestyle='--')
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.05])
plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.title('Receiver Operating Characteristic (ROC)')
plt.legend(loc="lower right")
plt.grid(True)
plt.show()

!pip install xgboost

Requirement already satisfied: xgboost in /usr/local/lib/python3.10/dist-packages (2.0.3)
Requirement already satisfied: numpy in /usr/local/lib/python3.10/dist-packages (from xgboost) (1.25.2)
Requirement already satisfied: scipy in /usr/local/lib/python3.10/dist-packages (from xgboost) (1.11.4)

from xgboost import XGBClassifier
xgb = XGBClassifier(n_estimators=100,random_state=42)
xgb_score=cross_val_score(xgb,X_train,y_train,cv=cv).mean()
xgb.fit(X_train, y_train)
# Predict probabilities for the test set
y_pred_prob = xgb.predict_proba(X_test)[:, 1]
xgb_y_pred = xgb.predict(X_test)
# Calculate the ROC curve
fpr, tpr, thresholds = roc_curve(y_test, y_pred_prob)
xgb_auc = roc_auc_score(y_test,y_pred_prob)

# Plot the ROC curve
plt.figure(figsize=(6, 6))
plt.plot(fpr, tpr, color='blue', label=f'XGBoost (AUC = {roc_auc_score(y_test, y_pred_prob):.2f})')
plt.plot([0, 1], [0, 1], color='red', linestyle='--')
plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.title('Receiver Operating Characteristic (ROC) Curve')
plt.legend(loc='lower right')
plt.grid(True)
plt.show()

from sklearn.linear_model import LogisticRegression as LR
# Train Logistic Regression model
logisticR = LR(max_iter=1000)
logisticR_score=cross_val_score(logisticR,X_train,y_train,cv=cv).mean()
logisticR.fit(X_train, y_train)

# Predict probabilities for the test set
y_pred_prob = logisticR.predict_proba(X_test)[:, 1]
logisticR_y_pred = logisticR.predict(X_test)
# Calculate the ROC curve
fpr, tpr, thresholds = roc_curve(y_test, y_pred_prob)
logisticR_auc = roc_auc_score(y_test,y_pred_prob)

# Plot the ROC curve
plt.figure(figsize=(6, 6))
plt.plot(fpr, tpr, color='blue', label=f'Logistic Regression (AUC = {roc_auc_score(y_test, y_pred_prob):.2f})')
plt.plot([0, 1], [0, 1], color='red', linestyle='--')
plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.title('Receiver Operating Characteristic (ROC) Curve')
plt.legend(loc='lower right')
plt.grid(True)
plt.show()

# 查看各模型classification_report
from sklearn.metrics import classification_report as CR
print('RandomForest'.center(50), CR(y_test,rdf_y_pred),sep='\n')
print('XGBoost'.center(55),CR(y_test,xgb_y_pred),sep='\n')
print('LogisticRegression'.center(50),CR(y_test,logisticR_y_pred),sep='\n')

                   RandomForest                   
              precision    recall  f1-score   support

           0       0.99      1.00      0.99      3428
           1       0.99      0.96      0.98      1072

    accuracy                           0.99      4500
   macro avg       0.99      0.98      0.98      4500
weighted avg       0.99      0.99      0.99      4500

                        XGBoost                        
              precision    recall  f1-score   support

           0       0.99      0.99      0.99      3428
           1       0.98      0.95      0.97      1072

    accuracy                           0.98      4500
   macro avg       0.98      0.97      0.98      4500
weighted avg       0.98      0.98      0.98      4500

                LogisticRegression                
              precision    recall  f1-score   support

           0       0.79      0.92      0.85      3428
           1       0.49      0.24      0.32      1072

    accuracy                           0.76      4500
   macro avg       0.64      0.58      0.59      4500
weighted avg       0.72      0.76      0.73      4500

score={'Model_score':[rdf_score,xgb_score,logisticR_score],'Auc_area':[rdf_auc,xgb_auc,logisticR_auc]}
score_com=pd.DataFrame(data=score,index=['RandomForest','XGBoost','LogisticRegression'])
score_com.sort_values(by=['Model_score'],ascending=False)

	satisfaction_level	last_evaluation	number_project	average_montly_hours	time_spend_company	Work_accident	left	promotion_last_5years	salary_num	department_num
count	14999.000000	14999.000000	14999.000000	14999.000000	14999.000000	14999.000000	14999.000000	14999.000000	14999.000000	14999.000000
mean	0.612834	0.716102	3.803054	201.050337	3.498233	0.144610	0.238083	0.021268	1.347290	5.870525
std	0.248631	0.171169	1.232592	49.943099	1.460136	0.351719	0.425924	0.144281	0.625819	2.868786
min	0.090000	0.360000	2.000000	96.000000	2.000000	0.000000	0.000000	0.000000	0.000000	0.000000
25%	0.440000	0.560000	3.000000	156.000000	3.000000	0.000000	0.000000	0.000000	1.000000	4.000000
50%	0.640000	0.720000	4.000000	200.000000	3.000000	0.000000	0.000000	0.000000	1.000000	7.000000
75%	0.820000	0.870000	5.000000	245.000000	4.000000	0.000000	0.000000	0.000000	2.000000	8.000000
max	1.000000	1.000000	7.000000	310.000000	10.000000	1.000000	1.000000	1.000000	2.000000	9.000000

	Model_score	Auc_area
RandomForest	0.990760	0.991931
XGBoost	0.986855	0.991493
LogisticRegression	0.768929	0.801972

Table of Contents ¶

Introduction¶

Data Inspection¶

Data Loading¶

Check for Missing Values¶

rename columns¶

View Data Information¶

Show Information of All Categorical Features¶

Categorical Digitization¶

Data Analysis¶

Descriptive Analysis¶

Employee Turnover Ratio¶

Relationship Between Employee Satisfaction and Years Spent at Company¶

Comparison of Employee Turnover by Department¶

Employee Salary Levels by Department¶

Impact of Employee Salary on Turnover Rate¶

Comparison of Employee Promotions in Last 5 Years and Turnover¶

Comparison of Employee Performance Evaluation and Turnover¶

Modeling¶

Correlation Analysis¶

Prepared training and testing sets¶

RandomForestClassifier¶

XGBoost¶

LogisticRegression¶

Summary¶

	satisfaction_level	last_evaluation	number_project	average_montly_hours	time_spend_company	left	sales	salary
0	0.38	0.53	2	157	3	1	sales	low
1	0.80	0.86	5	262	6	1	sales	medium
2	0.11	0.88	7	272	4	1	sales	medium
3	0.72	0.87	5	223	5	1	sales	low
4	0.37	0.52	2	159	3	1	sales	low